HyperIMTS: Hypergraph Neural Network for Irregular Multivariate Time Series Forecasting

Irregular multivariate time series (IMTS) are characterized by irregular time intervals within variables and unaligned observations across variables, posing challenges in learning temporal and variable dependencies. Many existing IMTS models either require padded samples to learn separately from temporal and variable dimensions, or represent original samples via bipartite graphs or sets.However, the former approaches often need to handle extra padding values affecting efficiency and disrupting original sampling patterns, while the latter ones have limitations in capturing dependencies among unaligned observations.To represent and learn both dependencies from original observations in a unified form, we propose HyperIMTS, a Hypergraph neural network for Irregular Multivariate Time Series forecasting.Observed values are converted as nodes in the hypergraph, interconnected by temporal and variable hyperedges to enable message passing among all observations.Through irregularity-aware message passing, HyperIMTS captures variable dependencies in a time-adaptive way to achieve accurate forecasting. Experiments demonstrate HyperIMTS's competitive performance among state-of-the-art models in IMTS forecasting with low computational cost.Our code is available at https://github.com/qianlima-lab/PyOmniTS.

Many real-world datasets, like medical records or sensor readings, track multiple measurements over time—but these measurements aren’t always taken at regular intervals or aligned with each other. This makes it hard for traditional models to comprehensively learn underlying patterns, as they often rely on fixed time steps or padding (adding artificial data), which can slower the processing speed and distort results. To tackle the problem, we developed HyperIMTS, a new method that treats each observed data point as part of a flexible network (a hypergraph), where connections capture both time and variable relationships between different measurements. Instead of forcing data into rigid structures, our model adapts to irregular gaps and mismatched observations, learning dependencies comprehensively and efficiently. Our approach improves forecasting accuracy for real-world irregular multivariate time series data—like predicting patient health trends—without heavy computational costs.